COSMOS-DSL: COmpoSable MOtion Specifications DSL using Popov-Vereshchagin Hybrid Dynamics solver

OS

• The code is tested on Ubuntu 20 and 22 versions.

Dependencies:

• rdflib
• jinja2
• urdfdom
• kdl_parser
• orocos_kinematics_dynamics
• gnuplot
• glfw3

Setup

Step 1

• clone the repository into a workspace

  mkdir -p ~/cosmos_dsl/src && cd ~/cosmos_dsl/src
  
  git clone --recurse-submodules -j8 https://github.com/vamsikalagaturu/cosmos-dsl.git .

Step 2

• You can install miniconda from here to create a virtual envrironment.

• Create a virtual environment from the cosmos_env.yml file.

  cd ~/cosmos_dsl/src
  
  conda env create -f cosmos_env.yml -n <name>

• Activate the environment

  conda activate <name>

Step 3

• Install the dependencies: kdl_parser, urdfdom, gnuplot, glfw3, yaml-cpp

  sudo apt-get install libkdl-parser-dev liburdfdom-dev libyaml-cpp-dev libglfw3 libglfw3-dev libgnuplot-iostream-dev

• Run this command to add the mujoco path to the ~/.bashrc file.

  echo -e 'export LD_LIBRARY_PATH=$HOME/.mujoco/mujoco210/bin
  export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/lib/nvidia 
  export PATH="$LD_LIBRARY_PATH:$PATH" 
  export LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libGLEW.so' >> ~/.bashrc

How to use

Step 1

• Install the modelling_tools package

  cd ~/cosmos_dsl/src/modelling_tools/
  
  (venv_name) pip3 install -e .

Step 2

• Convert the DSL to C++ code

  cd ~/cosmos_dsl/src/modelling_tools/
  
  (venv_name) python3 modelling_tools/move_arm_converter.py -d

• The converted C++ code is written to ~/workspace/src/arm_actions/src/ directory.

• Flags:

  -d         Print debug information
  -dg        Print rdf graph and the debug information
  -nr        Do not render the template 

Step 3

• Make and build the packages

  cd ~/cosmos_dsl/src/
  
  ./build.sh

• The output executables are written to ~/cosmos_dsl/outputs/ directory.

• Run the executables:

  cd ~/cosmos_dsl/outputs/
  
  ./arm_actions/<executable_name>

Same demo of end-effector moving down with constant velocity